一种基于趋势单产和遥感修正模型的混合估产模型

发布时间：2018-10-13 10:12

【摘要】：【目的】在分析国内外农作物估产方法的相关研究进展基础上,将传统统计估产方法和遥感估产方法相结合,提出一种新的混合估产模型。【方法】该模型由趋势单产、遥感修正单产和随机误差项三部分组成,其中趋势单产利用历史长时间序列的单产统计数据,通过多项式回归的方法结合ARIMA模型修正得到,遥感修正单产利用3个作物关键生育期NDVI和实测单产多元回归得到。为验证所提出估产方法的可行性和精度,利用2015年冬小麦关键生育期的三景环境卫星遥感影像和冬小麦实测地块单产数据以及近30年(1985—2014年)北京市各区县的冬小麦单产数据,对2015年的北京市的冬小麦单产进行估算,与真实值(2015年单产统计数据)对比。【结果】混合估产模型对北京市的冬小麦单产预测精度达到98.7%,各区县估产精度均超过90%,除房山(90.3%)外,各县单产预测相对精度均超过95%;传统趋势单产模型对北京市的冬小麦单产预测精度达到94.75%,但在区县尺度上,传统估产模型预测精度较低,对房山区的估产精度不足80%;引入ARIMA模型可以提高传统趋势单产模型的精度。修正后的趋势单产模型冬小麦单产预测精度平均提高了1.59%。本文建立的遥感修正模型,利用三景遥感影像修正结果最优,此方法使冬小麦估产精度整体提升3.55%,尤其是房山、平谷等区县,精度明显提升。【结论】该模型在市级尺度和县级尺度上预测冬小麦单产均取得较高精度,充分考虑冬小麦时间尺度和空间尺度上的变化,对农作物估产有一定的指导意义。
[Abstract]:[objective] on the basis of analyzing the related research progress of crop yield estimation methods at home and abroad, a new mixed yield estimation model was proposed by combining the traditional statistical yield estimation method with remote sensing estimation method. [methods] the model is based on the trend of yield per unit yield. There are three parts of modified unit yield and random error term by remote sensing, in which the trend unit yield is based on the statistical data of unit yield of historical long time series, which is obtained by polynomial regression method and ARIMA model correction. The modified yield of remote sensing was obtained by using NDVI of three crop key growth periods and multivariate regression of measured yield. To verify the feasibility and accuracy of the proposed yield estimation method, Based on the satellite remote sensing images of three scenes of winter wheat in the key growth period of winter wheat in 2015, the data of winter wheat yield per unit yield of winter wheat plots measured in the past 30 years (1985-2014) and winter wheat yield per unit yield data of every district and county in Beijing were used. To estimate the winter wheat yield in Beijing in 2015, [results] the prediction accuracy of mixed yield estimation model for winter wheat yield in Beijing reached 98.7%, and the precision of estimating yield in all districts and counties exceeded 90%, except Fangshan (90.3%). The relative precision of forecasting yield per unit yield of each county is more than 95. The precision of forecasting winter wheat yield per unit yield of Beijing by traditional trend yield model is 94.75, but on the scale of district and county, the precision of traditional yield estimation model is low. The yield estimation accuracy of Fangshan area is less than 80%, and the precision of traditional trend yield model can be improved by introducing ARIMA model. The prediction accuracy of winter wheat yield per unit yield was improved by 1.59 on average. The model of remote sensing correction established in this paper is optimized by using three scenes remote sensing images. This method improves the precision of winter wheat yield estimation by 3.55 as a whole, especially in Fangshan, Pinggu and other districts, counties and counties, such as Fangshan, Pinggu and so on. [conclusion] the prediction of winter wheat yield per unit yield on the scale of city and county level achieved high precision, and the changes of time scale and space scale of winter wheat were fully taken into account, which has certain guiding significance for crop yield estimation.
【作者单位】：北京师范大学地表过程与资源生态国家重点实验室;北京师范大学遥感科学与工程研究院;北京师范大学地理科学学部;北京市统计局;
【基金】：国家自然科学基金(41401479)
【分类号】：S127;S512.11

【相似文献】